Hidden Anomalies Detection in Large Arrays of Nuclear Power Plant Operating Data
Abstract
In this paper, it is investigated nuclear power plant operating data which was obtained from reactor main coolant pumps (MCP) of the third isolated generating plant of Kalinin NPP. It is necessary permanent monitoring for state of all pump components since breakdown of a reactor coolant pump leads to substantial economic losses. It is installed over 50 sensors of different control systems at the every MCP. Received data is stored but it is not analysed for the purpose of discovering joint dependencies between equipment pieces and unobvious, hidden trends of accident propagation. In this work, it was proposed techniques for detection of hidden anomalies and MCP operating regularity based on factor analysis, clustering and linear regression models. It was written a Python script which automates necessary calculations.
References
[1] Подготовка данных для проведения диагностики состояния ГЦН 3-го блока Калининской АЭС. М.Р. Лапшин, С.Т. Лескин, А.О. Скоморохов. ИАТЭ НИЯУ МИФИ, г. Обнинск.
[2] ОКБ «ГИДРОПРЕСС», 7-я международная научно-техническая конференция «Обеспечение безопасности АЭС с ВВЭР» ДИАГНОСТИКА ГЦН ВВЭР-1000 ПО ДАННЫМ ОПЕРАТИВНО-ТЕХНОЛОГИЧЕСКОГО КОНТРОЛЯ. С.Т. Лескин, В.И. Слободчук, А.С. Шелегов, М.Р. Лапшин, Обнинский Институт Атомной Энергетики, ИАТЭ НИЯУ МИФИ.
[3] R Core Team. 2015. R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing. https://www.R-project.org.
[4] Pedregosa, F., G. Varoquaux, A. Gramfort, V. Michel, B. Thirion, O. Grisel, M. Blondel, et al. 2011. “Scikit-Learn: Machine Learning in Python.” Journal of Machine Learning Research 12: 2825–30.
[5] Fischler, Martin A, and Robert C Bolles. 1981. “Random Sample Consensus: A Paradigm for Model Fitting with Applications to Image Analysis and Automated Cartography.” Communications of the ACM 24 (6). ACM: 381–95.
[6] Rousseeuw, Peter J, and Mia Hubert. 2011. “Robust Statistics for Outlier Detection.”Wiley Interdisciplinary Reviews: Data Mining and Knowledge Discovery 1 (1). Wiley Online Library: 73–79.
[7] Krewinkel, A., & Winkler, R. (2016). Formatting Open Science: agile creation of multiple document types by writing academic manuscripts in pandoc markdown (No. e2648v1). PeerJ Preprints.
[8] MacFarlane, J. (2013). Pandoc: a universal document converter. URL: http://pandoc.org.